Fig. 1

Overview of the PureCLIP approach. a PureCLIP starts with mapped reads from a target iCLIP/eCLIP experiment and derives two signals: the pulled-down fragment density and individual read start counts. Based on these two observed signals, it infers for each position the most likely hidden state. The goal is to identify all sites with an enriched + crosslinked state. Individual crosslink sites can then be merged to binding regions. b Additionally, information from input control experiments can be incorporated. Its fragment density is used to correct for a non-specific background signal, which reduces the number of false calls. c Furthermore, PureCLIP can incorporate information about CL motifs to reduce false calls caused by non-specific crosslinks. CL crosslink-associated